Underwater object detection using side-scan sonar (SSS) remains a significant challenge in marine exploration, especially for small objects. Conventional methods for small object detection face various obstacles, such as difficulties in feature extraction and the considerable impact of noise on detection accuracy. To address these issues, this study proposes an improved YOLOv11 network named YOLOv11-SDC. Specifically, a new Sparse Feature (SF) module is proposed, replacing the Spatial Pyramid Pooling Fast (SPPF) module from the original YOLOv11 architecture to enhance object feature selection. Furthermore, the proposed YOLOv11-SDC integrates a Dilated Reparam Block (DRB) with a C3k2 module to broaden the model’s receptive field. A Content-Guided Attention Fusion (CGAF) module is also incorporated prior to the detection module to assign appropriate weights to various feature maps, thereby emphasizing the relevant object information. Experimental results clearly demonstrate the superiority of YOLOv11-SDC over several iterations of YOLO versions in detection performance. The proposed method was validated through extensive real-world experiments, yielding a precision of 0.934, recall of 0.698, mAP@0.5 of 0.825, and mAP@0.5:0.95 of 0.598. In conclusion, the improved YOLOv11-SDC offers a promising solution for detecting small objects in SSS images, showing substantial potential for marine applications.

Joint segmentation of heterogeneous side-scan sonar datasets requires learning from different acoustic distributions and annotation scopes. Dataset-aware masked cross-entropy attenuates classes outside each dataset’s annotation scope, potentially reducing incorrect negative supervision while weakening cross-class competition. This study evaluates this trade-off using AI4Shipwrecks, MCM, and SubPipe in a shared four-class space. Experiments compare unmasked cross-entropy, soft attenuation, and hard exclusion across input representations and sampling policies. With five-channel SegFormer-B2, soft DMCE achieved a mIoU of 0.4908, below unmasked cross-entropy at 0.5234 in all five matched seeds; their foreground macro-IoU values were 0.0409 ± 0.0360 and 0.0694 ± 0.0413, respectively. After averaging the two repeated runs within each seed, hard masking achieved a descriptive macro mIoU of 0.1544 ± 0.0670 and a foreground macro IoU of 0.0144 ± 0.0153 across five seed-level summaries. Under PhysD input, the mean ordering reversed: soft DMCE achieved 0.5204 versus 0.4952 for cross-entropy. Dataset-balanced sampling increased cross-entropy performance to a macro mIoU of 0.5505 and a foreground macro IoU of 0.1138 ± 0.0144. Pipeline IoU nevertheless remained near zero despite SubPipe mIoU approaching 0.5, and held-out foreground IoU was zero when the only positive supervision for a category was excluded. Overall, the effect of attenuation varied across training configurations. These results show that background-inclusive macro mIoU should be interpreted alongside foreground-specific metrics when assessing joint segmentation performance.